HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Stauffer, P. H. Articles by Hopkins, J. K. Search for Related Content PubMed Articles by Stauffer, P. H. Articles by Hopkins, J. K. GeoRef GeoRef Citation Agricola Articles by Stauffer, P. H. Articles by Hopkins, J. K. Related Collections Volatile Organic Compounds Solute Transport Models Diffusion

SPECIAL SECTION: LOS ALAMOS NATIONAL LABORATORY

Vadose Zone Transport of 1,1,1-Trichloroethane

Conceptual Model Validation through Numerical Simulation

^a Geology, Hydrology, and Geochemistry Group EES-6, Earth and Environmental Sciences Division, Los Alamos National Laboratory, Los Alamos, NM 87545
^b Environmental Geology and Spatial Analysis Group EES-9, Earth and Environmental Sciences Division, Los Alamos National Laboratory, Los Alamos, NM 87545
^c Environmental Characterization and Remediation Group RRES-ECR, Risk Reduction and Environmental Science Division, Los Alamos National Laboratory, Los Alamos, NM 87545
^* Corresponding author (stauffer{at}lanl.gov)

Received 11 August 2004.

	ABSTRACT

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
A conceptual model is developed to better understand vadose zone vapor-phase diffusion within the mesas of the Pajarito Plateau at Los Alamos National Laboratory. We focus on 1,1,1-trichloroethane (TCA) vapor transport from a liquid-waste disposal site. The conceptual model incorporates several physical processes, including partitioning of TCA into the liquid phase, saturation-dependent diffusion, diffusion through asphalt, and interaction with the atmosphere. Three-dimensional numerical simulations that use the conceptual model of TCA transport are then calibrated to pore-gas monitoring data. Adjustable parameters in the numerical simulations are limited to (i) the vapor-phase diffusion coefficients for the different geologic units, asphalt cover, and land–atmosphere boundary layer and (ii) the fixed concentrations in the two shaft clusters. By including all of the components of the conceptual model in our numerical simulations we were able to achieve a reasonable match between the simulated plume and site data for two alternate conceptual models of asphalt, one with asphalt as a diffusive barrier and one without asphalt as a diffusive barrier. A goodness-of-fit analysis shows that the best-fit simulations are highly correlated to 132 data points from 21 boreholes. The simulations demonstrate that diffusive behavior describes the general characteristics of the current subsurface vapor plume. Effective vapor-phase diffusion coefficients used in the simulations that best fit the data suggest that barometric pumping is not contributing to diffusion in the deep vadose zone; however, it is likely that barometric pumping is occurring in fractures near the mesa edge. We conclude that asphalt is most likely acting as a barrier to diffusion at this site. The numerical simulations validate that the conceptual model developed for this study is a useful tool for analyzing TCA transport within the mesas of the Pajarito Plateau.

Abbreviations: B&K, Bruel and Kjaer • COPC, chemicals of potential concern • DEM, digital elevation model • MDA, Material Disposal Area • TCA, 1,1,1- trichloroethane • VOC, volatile organic compound

	INTRODUCTION

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
MIGRATION OF VAPOR-PHASE contaminants in the vadose zone is often much more rapid than migration of those in the liquid phase because vapor diffusion is several orders of magnitude larger than liquid diffusion for the same molecule (Fetter, 1999). Because of their high vapor pressure and low solubility in water, volatile organic compounds (VOCs) have a large potential to migrate in the vadose zone (Jury et al., 1990). The difference is accentuated as the saturation of the rocks approaches residual moisture, at which point movement in the liquid phase is essentially stopped (Conca and Wright, 1992). As pore spaces become drier, air pathways become better connected and vapor can more easily migrate under molecular diffusion or pressure driven flow. Rapid vapor transport can also be caused by large variations in barometric pressure (Auer et al., 1996) and pressure gradients resulting from topographic features (Weeks, 2001).

Several factors can act to reduce the ability of vapor-phase VOCs to migrate in the vadose zone. As pore spaces become saturated, vapor-phase tortuosity increases and both the effective vapor-phase diffusion coefficient and the gas-phase permeability are reduced (Jury et al., 1991). Volatile organic compounds tend to have low solubility, but some partitioning into vadose zone water occurs and can slow the migration of VOC vapor (Jury et al., 1990). Partitioning is often assumed to be an equilibrium phenomenon; however, Thomson et al. (1997) reported that this may be an oversimplification of field conditions. Another process that slows VOC migration is sorption onto mineral surfaces, which may or may not be reversible. Slow desorption of VOC in the vadose zone can lead to long-term sources that are extremely difficult to remediate. Finally, degradation of VOCs by both abiotic and biotic processes can help to attenuate a contaminant plume (Vogel et al., 1987). However, abiotic degradation can be quite slow, and biotic degradation can be slow and inefficient in the vadose zone where liquid nutrient flux and waste removal are impeded by low relative permeability of liquids caused by the presence of soil gas in the pore structure.

The purpose of this work is to develop a conceptual model (Birdsell et al., 2005) of subsurface VOC transport and to validate the conceptual model through numerical simulations of TCA transport in the vadose zone at Los Alamos National Laboratory. The simulations are calibrated to site data from a network of vapor-phase monitoring ports located in boreholes that surround the plume. Adjustable parameters in the simulations are limited to (i) the vapor-phase diffusion coefficients for the different geologic units, asphalt cover, and land–atmosphere boundary layer and (ii) fixed concentrations of TCA in the source region. The modeling allows the relative importance of different transport processes to be assessed and helps to show which processes are dominant in our conceptual model. A valid conceptual model for VOC transport is crucial for predicting future risk associated with the site and is a necessary step in gaining regulatory approval for site closure (National Research Council, 2001). Once validated, the conceptual model, as implemented in the numerical simulations, can be used with some confidence to simulate future plume behavior with respect to remediation options.

	SITE DESCRIPTION

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Geologic Setting
Material Disposal Area (MDA) L is located on Mesita del Buey, a narrow finger mesa situated near the eastern edge of the Pajarito Plateau (Broxton and Vaniman, 2005; McLin et al., 2005). The Pajarito Plateau is located on the eastern flank of the Jemez volcanic center, and the rocks that form the plateau were created in two main ignimbrite eruptions that occurred at approximately 1.61 and 1.22 Ma (Izett and Obradovich, 1994). Subsequently, the plateau has been incised by canyons that drain to the Rio Grande. The regional aquifer is located approximately 300 m below the surface of MDA L, and no perched water was encountered during drilling at the site.

Figure 1 gives the general geography for MDA L and the surrounding area, and Fig. 2 shows the approximate site stratigraphy on a north–south cross section of the mesa with several of the boreholes from Fig. 1 projected onto the plane. The upper five stratigraphic units make up the Tshirege Member of the Bandelier Tuff. The Bandelier Tuff is composed of nonwelded to moderately welded rhyolitic ash-flow and ash-fall tuffs interbedded with thin pumice beds. Qbt 2 and Qbt 1vc contain nearly ubiquitous, vertical cooling joints (Purtymun and Kennedy, 1971), with average joint spacing in Qbt 2 varying from 0.9 to 3.3 m (McLin et al., 2005). Joints are typically <6 mm, but some openings are as large as 50 mm (Neeper and Gilkeson, 1996). Clay fills many of the joints to depths of 6 m, but below this depth joints are either open or filled with powdered tuff (Neeper and Gilkeson, 1996). The vertical joints in Qbt 1vc are almost all open and provide rapid equilibration of pressure changes during pumping tests (Neeper 2002). More discussion of the fracture characteristics and petrography of the Bandelier Tuff on Mesita del Buey can be found in McLin et al. (2005). The base of the Tshirege Member is the Tsankawi pumice (Qbt t) that is typically between 1 and 2 m thick. The Cerro Toledo interval (Qct) is comprised of volcanoclastic sediments interbedded with minor pyroclastic flows, and separates the Tshirege and Otowi Members (Qbo) of the Bandelier Tuff. The Otowi Member is nonwelded to poorly welded and is not fractured (Vaniman et al., 1996). The basal subunit of the Otowi member is the Guaje Pumice. The Cerros del Rio Basalt (Tb4), which comprises at least 35% of the vadose zone, displays wide variability (Turin, 1995), ranging from extremely dense with no effective porosity, to highly fractured, to so vesicular as to appear foamy. Neeper (2002) reported that the pressure signal within the basalt (Location G on Fig. 2) is nearly in phase with the atmosphere and that pressure data from Boreholes 1015 and 1016 (Fig. 1) can be explained by extremely high air permeability (1 x 10^–9 m²) connected to a distant (1.5 km) outcrop. The Puye Formation, a conglomerate of volcanic cobbles and boulders in a matrix of sand, silt, and clay, underlies the Cerros del Rio Basalts and extends from the base of the vadose zone well into the saturated zone. Material properties relevant to the conceptual model of TCA transport are discussed below. The geology of the Pajarito Plateau was described in more detail in Broxton and Vaniman (2005), Broxton and Reneau (1995), and Reneau et al. (1998).

View larger version (47K): [in this window] [in a new window]   Fig. 1. Geographical information for MDA L and the surrounding area.

View larger version (24K): [in this window] [in a new window]   Fig. 2. Two-dimensional site stratigraphy (from Neeper, 2002).

One pit, three surface disposal impoundments, and 34 disposal shafts are the Potential Release Sites (PRS) at MDA L (Fig. 3) (LANL, 2002). These PRSs had varying purposes and were used for different time periods. The disposal pits are not considered in this study because they present no viable source for VOC in the subsurface. Disposal shaft numbers 1 through 28 operated from 1975 through 1985, while shaft numbers 29 through 34 operated from 1983 through 1985. After 1985, most of the 2.5 acres comprising MDA L were covered with asphalt on which temporary storage facilities for chemical waste were built. There is little information on specific chemicals, timing, or quantities of waste that were disposed of at MDA L. However, the major COPCs measured at this site are found in pore-gas sampling and include a host of VOCs (LANL, 2002). The 34 shafts received metal drums (200 L) containing chemical liquid waste. The waste drums were packed in lifts with one to six barrels per layer. In the shafts, layered waste was covered with crushed tuff to provide absorbent material as well as structural support for the drums. Additionally, unknown quantities of small containers and free liquids were dropped directly into the shafts. The locations of the pits and shafts can be seen in Fig. 3.

View larger version (23K): [in this window] [in a new window]   Fig. 3. Location map for waste disposal shafts and pits at MDA L.

Figure 4 shows two quarters of TCA monitoring representing the largest and smallest plume measured from 1997 to 2003. In this figure the minimum contour of 10 mL m^–3 represents the lower limit of data reliability, which was reported as 5 to 10 mL m^–3. The major vapor-phase contaminant measured in the plume is 1,1,1-TCA, which comprised approximately 75% by volume of the spatially averaged plume. The second most prevalent VOC found was TCE, comprising 12.5% by volume of the plume, while Freon 113 comprised 11.2% by volume of the averaged plume. These values are averaged over 140 sampling locations measured using the B&K field-screening method. Although these numbers are spatial averages, individual sampling ports showed wide variation in the ratios of the most prevalent VOCs, and the more complete analytic laboratory analyses showed some ports (e.g., Well 54-2032 at 47.5 m below the collar) have significant percentages of compounds such as naphthalene (12%) and 1,1-dichloroethene (14%) (Smith et al., 1998, 1999a, 1999b).

View larger version (61K): [in this window] [in a new window]   Fig. 4. Material Disposal Area L 1,1,1- trichloroethane concentration data contoured on cross sections and in map view.

	CONCEPTUAL MODEL

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Our conceptual model for TCA transport at this site contains several key features including saturation-dependent vapor diffusion, equilibrium partitioning between TCA in the liquid and gas phases, topographic boundaries in contact with the atmosphere, asphalt covering part of the site, and reduction in diffusion across the land–air interface due to boundary layer processes that may impede mass transfer. We simplify the analysis by considering only transport of TCA, which is justified by the fact that this VOC is consistently measured in the pore gas at concentrations several times higher than other components of the plume. However, other constituents can readily be incorporated by changing the physical properties listed in Table 2 to appropriate values from the literature. Diffusive processes are assumed to be the fundamental drivers controlling the migration of TCA at the site. Vapor-phase TCA diffusion coefficients have not been measured in Bandelier Tuff; however, TCA and TCE have similar molecular weights, and we assume that their in situ diffusion coefficients will be quite similar (Table 2).

Vapor-phase diffusion depends on water saturation in our conceptual model, as suggested by the work of Millington (1959). The three upper Tshirege units (Fig. 2) are extremely dry, with saturations ranging from approximately 4 to 15%. The underlying Cerro Toledo and Otowi Members have higher saturations, in the 30 to 40% range. Because the saturation-dependent diffusion coefficient, D_v(sat) is inversely related to the saturation (Conca and Wright, 1992), diffusive flux is expected to be greater in the upper units than in the Cerro Toledo and Otowi Member.

In our conceptual model, the effective vapor-phase diffusion coefficient accounts for both vapor diffusion and diffusive spreading induced by barometric pumping. The primary effect of barometric pumping on chemical vapors is to increase the apparent rate of vapor diffusion because the air movement in the subsurface has no net velocity and is merely shifted back and forth, which leads to increased spreading of tracer (Auer et al., 1996; Neeper, 2001, 2002). For this reason, the values we assign for D_v^* in each geologic unit or material are guided by laboratory measured values (Fuentes et al., 1991) and the work of Millington (1959) but remain adjustable parameters in our conceptual and numerical models. Additionally, barometric pumping can propagate much further in a fractured system than in a nonfractured porous medium and has been proposed as a mechanism for increased apparent water diffusion beneath MDA L (Neeper and Gilkeson, 1996; Neeper 2001, 2002). The effects of such fracture-induced barometric pumping are implicitly included in our conceptual model as part of the effective vapor-phase diffusion coefficient, and the numerical simulations are used to determine if increased diffusion caused by barometric pumping is necessary to match the pore-gas monitoring data.

Water saturation in the rock also plays a role with respect to partitioning between the liquid and gas phases. We assume that the partitioning of TCA between the vapor phase and the pore water can be described by Henry's Law. Henry's Law assumes equilibrium and implies that the partitioning between phases is fast relative to the diffusive transport flux (Fetter, 1999). Low water saturation in Qbt 2 allows for very little storage of VOCs in the liquid phase. Because of higher saturations, more partitioning of TCA into the liquid phase occurs in the Cerro Toledo and Otowi, retarding the diffusion of TCA through these units.

MDA L is located on the edge of a narrow mesa, so the interaction of the subsurface vapor-phase TCA with atmospheric air must be considered. The topographic relief of the mesa provides an atmospheric boundary condition of zero concentration where the plume intersects the mesa top and sides (LANL, 1994). The zero-concentration atmospheric boundary is essentially an infinite sink that can accept the TCA vapor that diffuses outward from the high concentrations found in the source region. This boundary maintains a steep concentration gradient between the source and the mesa sides, which limits plume growth both along the axis of the mesa and to depth. Reduction in diffusion across interfaces can be caused by increased tortuosity in alluvium, soils that contain higher clay and silt fractions than intact bedrock, and boundary layer effects (Jury et al., 1990), and for this reason, we examine the sensitivity of the model to a decrease in the D_v^* across the land–air interface.

Another factor that potentially affects the ability of TCA vapor to migrate upward into the atmosphere is the layer of asphalt that was applied to the surface of MDA L. Because we do not have measurements of flux across the asphalt, the behavior of this layer is poorly constrained. We present two alternate conceptual models for the asphalt. The first envisions the asphalt as a nearly perfect barrier to diffusion. The second conceptual model assumes that the asphalt is ubiquitously cracked, providing no barrier to vapor-phase TCA diffusion.

The migration of TCA vapor from the barrels in the shafts is conceptualized as a time-release phenomenon. This is based on the idea that liquid will leak slowly and quickly volatilize. Sudden increases of TCA concentrations in the source region are possible if individual drums of solvents burst; however, barrel corrosion data suggest slow leakage to be a more plausible scenario (Lyon et al., 1996). Additional evidence presented below supports this conceptual model.

Processes Not Included in the Conceptual Model
In low permeability rock (<1 x 10^–14 m²) vapor-phase diffusion in porous media may be better modeled by equations such as the Dusty Gas Model (Webb, 1998). However, permeabilities at this site are >1 x 10^–13 m², so we did not consider this theory as part of our conceptual model.

Flow of liquid water can affect movement of vapor in the subsurface. At this site, however, the estimated infiltration is approximately 1 mm yr^–1 or less (Kwicklis et al., 2005; Birdsell et al., 2000). This low rate of flow will have very little effect on the transport of TCA, and we did not consider the movement of liquid water as part of our conceptual model.

Thermal diffusion is very slow in the earth, and daily to weekly variations will propagate on the order of 1 m or less (Hillel, 1982). Seasonal variation may cause long wavelength temperature oscillation in the upper few meters; however, the temperature changes will be on the order of a few degrees. Below 3 to 4 m depth the expected temperature of the subsurface should remain nearly constant (Hillel, 1982). Therefore, variations in temperature were not considered in this current study, and we assumed that the yearly average temperature is adequate to define the state of the mesa.

Pressure changes caused by large-scale atmospheric disturbances, thermal gradients, and wind effects have been shown to induce advective flow in the vadose zone (Weeks, 2001; Neeper and Gilkeson, 1996). Calculations in Stauffer and Rosenburg (1999) show that the topographic effect is unimportant because of the low thermal gradient and limited topographic relief of the mesa-canyon systems limit advective upward flow to about 2 cm yr^–1. As described above, the effects of barometric pumping are indirectly included as part of D_v^*, and we do not attempt to simulate daily pressure variations in the simulations. Prevailing wind could create lower pressure on the downwind side of the mesa, leading to preferential advection. Because no such preferential shift is seen in the pore-gas monitoring data we assumed that wind effects can also be neglected. Therefore, our conceptual model includes no variation in pressure and no advective vapor-phase flow. Because we did not see any evidence of preferential advective air flow, we did not explicitly include fractures in our conceptual model.

Sorption of TCE onto unsaturated Bandelier tuff was measured by Ong and Lions (1991). Results from their study show that at high vapor-phase concentrations (20000 mL m^–3) and low moisture contents (<0.3%), some amount of TCE partitions onto the solid mineral surfaces. Given the much lower TCA concentrations found at MDA L (<3000 mL m^–3) and the range of in situ saturations found within the mesa beneath MDA L, their data suggest that sorption should be minimal relative to liquid-phase partitioning; therefore, sorption was not considered as part of the conceptual model.

Although we cannot eliminate the possibility that liquid TCA existed at some point in time within the pore space that surrounds the shafts, no liquid TCA was found in any of the boreholes drilled under the site (Fig. 1). Furthermore, at an average mesa-top internal air pressure of approximately 800 mb, the vapor pressure of 1,1,1-TCA (Table 2) would lead to an equilibrium vapor concentration >160000 mL m^–3. Because measured concentrations are more than an order of magnitude below the 1,1,1-TCA vapor–liquid equilibrium concentration, the vapor monitoring data provide additional evidence that liquid TCA is not present. Therefore, transport of liquid TCA were not included in the conceptual model.

	NUMERICAL MODEL

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
The simulations used to validate the conceptual model are performed using FEHM, a three-dimensional finite-volume heat and mass transfer code suitable for simulating systems with complicated geometries (Zyvoloski et al., 1997). The governing equations in FEHM arise from the principles of conservation of water mass, air mass, contaminant mass, and energy. The advection–dispersion equation that governs solute transport in FEHM reduces to the diffusion equation under no-flow conditions (Fetter, 1999; Zyvoloski et al., 1997).

Application of the Conceptual Model to the Numerical Model
In the following sections we describe how the details of the conceptual model are implemented in the numerical simulations. The simulations implement the processes and features discussed in the conceptual model in conjunction with constraints on transport parameters suggested by measurement data. The numerical model is then calibrated by adjusting (i) the vapor-phase diffusion coefficients for the different geologic units, the asphalt cover, and the land–atmosphere boundary layer and (ii) the fixed concentrations of TCA in the source region to create a numerical model of the site that fits a majority of the pore-gas sampling data. The simulations use a simplified site history that includes estimates of the timing and location of waste emplacement and asphalt emplacement.

Model Domain and Computational Grid
The MDA L site model is a three-dimensional representation of the subsurface stratigraphy including the surface topography. The model domain covers a rectangular map area that is considerably larger than the MDA L site boundary (Fig. 5) . The grid is 411 m wide (east–west) and 290 m wide (north–south). The land surface in the model domain is based on digital elevation model (DEM) data, which allows the topography of the mesa–canyon system to be captured. The grid uses a subset from the DEM data to approximate the surface with 15-m spacing. The model surface shown in Fig. 5 is comparable to the site topography seen in Fig. 1. Node spacing is 15.24 m in both the x and y directions and is variable in the z direction from a minimum spacing of 1 m to a maximum of 15.24 m.

View larger version (148K): [in this window] [in a new window]   Fig. 5. Model topography and computational grid.

View larger version (84K): [in this window] [in a new window]   Fig. 6. Model stratigraphy.

v

Water flow is restricted by using van Genuchten (1980) water characteristic curve parameters that result in no appreciable water movement. Vapor-phase advective flow is eliminated by fixing a horizontally stratified temperature gradient in the model domain that is below the critical Rayleigh number and results in no convective motion (Stauffer et al., 1997). Before running the contaminant transport simulations, a static air-pressure field is established by running the model until pressures and temperatures reach equilibrium. This ensures that the transport simulations are not affected by transient behavior associated with establishing a static air-pressure field. The steady-state initial condition has no TCA present and is meant to represent the mesa before the release of contaminants.

Hydrologic Parameters
Porosity in the simulations is fixed to the mean values reported by Springer (2005) for samples located near MDA L (Table 1). An important assumption inherent in all the simulations is that we fix the saturations in the subsurface to values within the measured range that tend to accentuate the effects of vapor–liquid partitioning and vapor-phase diffusion as a function of saturation. For example, measured saturations in Unit 2 range from 2 to 10% (Birdsell et al., 2000); however, for this study we fixed the saturation of this unit to the lowest measured value of 2%. Unit 1v and 1g were fixed at 15% saturation, and the Cerro Toledo and Otowi units were both fixed at the high end of measured values (40 and 35%, respectively). Deeper units play little role in transport at this site, and units below the Otowi were fixed at 2% saturation. Saturation values used in all the simulations are listed in Table 1.

Shaft Location, Source Release, and Asphalt Properties
The model node spacing is too coarse to explicitly include each shaft present at MDA L. Therefore, we group the shafts into two clusters, with the eastern shaft cluster representing Shafts 1 through 28 and the western shaft cluster representing Shafts 29 through 34. Each shaft cluster is specified using three nodes, which include a volume of the model domain (3520 m³) extending from 2 m below the surface to a depth of approximately 20 m.

TCA is introduced to the model shaft nodes based on a simplification of the available data. The shaft nodes are assigned fixed vapor concentrations of TCA based on the highest measured concentrations from vapor ports near the shafts. Because maximum measured concentrations could have varied through time, we simulated a range of fixed concentration sources (2500–18000 mL m^–3) in the shafts as part of the calibration process. Higher early TCA concentrations are justified because early disposal practices were less stringent.

Two end-member cases for the asphalt were explored. The first case envisioned the asphalt as a completely sealed layer with a diffusion coefficient that causes nearly all the TCA to be trapped, while in the second case we assigned the asphalt the same diffusion coefficient as Unit 2 of the Tshirege Member of the Bandelier Tuff (Table 3).

Model Calibration
Model calibration for the two end-member asphalt cases was performed manually by varying the adjustable parameters until the data–model correlation improved based on the goodness-of-fit analysis discussed in the following section. Parameters varied during the calibration were (i) D_v^* of the geologic units and land–air interface (Table 4) and (ii) the fixed TCA concentrations in the two shaft clusters. As discussed in the conceptual model section, D_v^* of the geologic units was guided by measurements but allowed to vary from two times higher to one-half times lower than the values suggested by the measurements of Fuentes et al. (1991), the in situ saturation, and the theory of Millington (1959). Manual calibration is somewhat laborious, but the process leads to increased appreciation for the interplay of the adjustable parameters and much greater familiarity with site-specific details.

For this study, we simulated growth of the MDA L TCA plume from 1975 to 2000. To compare the simulations to the data, we used a subset of the pore-gas monitoring data to generate an average plume that represents the year 2000. The average data plume used to calibrate the simulations was generated by averaging all measurements at each pore-gas sample port from the first quarter of 1999 through the last quarter of 2000. Due to inconsistencies in some of the sample data (i.e., sporadic null values in an otherwise strong signal), some individual values were removed from the analysis. In a few instances, data varied widely from quarter to quarter and entire ports were deemed unreliable and removed from the analysis. The final average data plume used in the calibration process consisted of 132 measurements from 21 wells that surround the site.

Goodness of Fit
A simple linear regression of the model–data concentration pairs was used to visualize the goodness of fit between the model results and the monitoring data (Fig. 7 and 8) . Log-space was used to display the regression because the data span three orders of magnitude, and the number of low concentration measurements far exceeds the number of high concentration measurements. Because the numerical grid is sparse and not every data sampling location corresponds to a node, the simulated borehole concentrations were computed through quad-linear interpolation to the nearest four grid points.

View larger version (21K): [in this window] [in a new window]   Fig. 7. Data–model regression for (A) the best-fit simulation with asphalt and (B) the best-fit simulation without asphalt. The details for each simulation are listed in Table 5.

View larger version (39K): [in this window] [in a new window]   Fig. 8. Data–model regression for sensitivity simulations. In each case, the simulation has been modified from the best-fit simulation. The details for each simulation are listed in Table 5.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Best Fit Simulation with Asphalt as a Diffusive Barrier
Figure 7A shows the data–model regression for the best fit that we achieved for the case where asphalt was simulated to be a diffusive barrier. For the remainder of this paper we refer to this simulation as Simulation A. This simulation contains all the processes discussed in the conceptual model, including liquid–vapor partitioning, diffusion as a function of saturation, interaction with the atmosphere, and a reduction in diffusion at the land–air interface. D_v^* used for the geologic units is 3 x 10^–6 m² s^–1 for Qbt 2, basalt, and Puye; 2 x 10^–6 m² s^–1 for Qbt 1v and Qbt 1g; and 5 x 10^–7 m² s^–1 for the Otowi and Cerro Toledo. The land–air D_v^* is set to 1 x 10^–6 m² s^–1, and the asphalt diffusion coefficient is set to 1 x 10^–14 m² s^–1 (Table 3). For this simulation, 18000 mL m^–3 TCA is fixed in the eastern shaft cluster from 1975 to 1983; then the concentration in this cluster is reduced to 4000 mL m^–3 from 1983 to 2000. The concentration of TCA in the west shaft cluster is fixed at 4000 mL m^–3 from 1983 to 1985, then reduced to 3000 mL m^–3 from 1985 to 2000 (Table 5). Simulation A results in approximately 1000 kg of TCA in the modeled plume in the year 2000, with approximately 260 kg of TCA in the soil moisture due to liquid-phase partitioning (Henry's Law).

Effective vapor-phase diffusion coefficients for this simulation are quite close to the measured TCE values from Fuentes et al. (1991) that we are using as analogs for TCA (Table 5). The fact that this simulation can fit the data using values on the low end of the measured analog diffusion coefficients implies that an increased rate of diffusion due to barometric pumping deep within the mesa, as proposed by Neeper and Gilkeson (1996), is not occurring. We note, however, that throughout the calibration process, simulated TCA concentrations in Wells 2021, 2023, and 2031 consistently overestimated the data. These wells lie close to the mesa edge where barometric pumping within exposed fractures on the mesa edge is most likely to have an impact on TCA concentration.

The mean weighted residual for this simulation is –0.028 and confirms that the data–model error is nearly evenly distributed about the "data = model" line (Table 6). The variance for this simulation is 0.103 and provides a measure of the scatter. Scatter increases at the lower end of concentration for two reasons. First, the quarterly data measurements have higher standard deviations at low values, and second, these points are the furthest from the shaft clusters and may be affected by secondary processes that are not included in our conceptual model, such as an increase in the rate of diffusion near the mesa edge caused by localized barometric pumping (Neeper, 2002).

v

v

The mean weighted residual for Simulation B is 0.018, showing that the data–model error is nearly evenly distributed about the "data = model" line (Table 6). The variance for Simulation B is 0.099, virtually identical to the variance of Simulation A. Although Simulation B has nearly the same goodness-of-fit parameters as Simulation A, Simulation B requires fixed source concentrations (6000 mL m^–3) that are well above the highest recent measurements taken near the source region (approximately 3900 mL m^–3). This leads us to conclude that Simulation A is the more likely of the two alternate conceptual models and that asphalt is acting as a diffusive barrier at MDA L.

Model Sensitivity
Finally, we discuss simulations that show how the goodness-of-fit of Simulation A is affected by changing or removing individual parameters or processes. Figure 8 shows data–model regression plots for a series of simulations (Simulations C–H), each of which makes a single change with respect to Simulation A. Table 6 lists mean normalized residual and variance for Simulations C through H, with a brief descriptor indicating what change has been made relative to Simulation A.

Simulation C shows how increasing D_v^* at the land–air interface effects the growth of the plume. When the land–air D_v^* is increased to 3 x 10^–6 m² s^–1, the plume does not grow as large and the cluster of points in the data–model regression shifts to the right of the "data = model" line.

Simulation D shows how decreasing D_v^* at the land–air interface affects the growth of the plume. When the land–air effective diffusion is decreased to 1 x 10^–7 m² s^–1, the plume grows much larger, which results in the data model regression points falling on the left side of the "data = model" line. Because simulated concentrations at monitoring points located in wells near the eastern shaft cluster (2002, 2014, 2089) are all shifted to values well above the data, this simulation shows that a reduction of the land–air D_v^* to 1 x 10^–7 m² s^–1 is unlikely on the mesa top surrounding MDA L.

Simulation E removes the process of partitioning of TCA into the liquid phase. Without liquid-phase partitioning, there is no storage for the TCA in the pore water, and the plume is able to grow larger, mainly at the boundaries where concentrations are below 100 mL m^–3. This simulation results in simulated TCA concentrations of >10 mL m^–3 in the basalt in Well 1016 where the data show values of near zero. The source region is not affected noticeably by removal of liquid-phase partitioning, and the data–model regression points above 100 mL m^–3 remain virtually unchanged. This is because the region near the source is closer to a true steady state, and storage becomes irrelevant to the final concentration gradient that will develop in a given system. This implies that the plume has not reached equilibrium and is continuing to grow away from the source. Although the plume may be continuing to grow spatially, evidence for this growth may be difficult to see in the monitoring data because the total amount of liquid-phase storage is proportional to the volume that the plume occupies, and incremental changes in the total plume radius must fill the storage of ever increasing volume shells.

Simulation F removes the dependence of effective diffusion on saturation and assumes that all the geologic units have an effective diffusion coefficient of 3 x 10^–6 m² s^–1. This simulation leads to much more mass (1270 kg) in the mesa than in Simulation A and greatly overestimates the data at nearly every sample location. This simulation also results in concentrations >10 mL m^–3 in the basalt in Wells 1015 and 1016 and 60 mL m^–3 in the Otowi Member in Borehole 1015 where the data show TCA concentration values near zero.

Simulation G increases the effective vapor-phase diffusion coefficient in each geologic unit by a factor of 1.5 and leads to model results that overestimate the data with a total TCA mass of approximately 1240 kg in the mesa. In the last sensitivity, Simulation H, the effective vapor-phase diffusion coefficient in each geologic unit is multiplied by a factor of 0.66. This simulation underestimates the data for nearly every measurement value, except in the source region where the high fixed values remain relatively unchanged for all the simulations shown on Fig. 8. Simulations G and H show that the numerical model is very sensitive to changes in the effective vapor-phase diffusion coefficient.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
A conceptual model of TCA transport in the vadose zone was successfully validated using three-dimensional numerical simulations. These simulations are able to achieve a reasonable match between the simulated plume and site data for two alternate conceptual models of asphalt, one with asphalt as a diffusive barrier and one without asphalt as a diffusive barrier. A goodness-of-fit analysis showed that both conceptual models of asphalt can be correlated to 132 data points from 21 boreholes. The conceptual model that did not include the asphalt as a barrier required higher simulated source concentrations than have been measured, and from this we conclude that asphalt is most likely acting as a barrier to diffusion at this site.

Sensitivity analyses demonstrate that all of the processes included in the conceptual model are important for describing the current subsurface vapor plume. Effective vapor-phase diffusion coefficients required to fit the data suggest that, in the deep vadose zone, barometric pumping is not contributing to the diffusion. In contrast, within fractures on the mesa edge, we conclude that barometric pumping may be occurring. The simulations confirm that the working conceptual model developed for this study is a useful tool for analyzing TCA transport within the mesas of the Pajarito Plateau.

	ACKNOWLEDGMENTS

 
This work was supported by the LANL Environmental Restoration Project for the U.S. Department of Energy. The authors thank Veraun Chipman for his help with data visualization. We thank Don Neeper for invaluable insight and help with many aspects of this analysis. We thank Andrew Wolfsberg, David Rogers, Bryan Travis, Diana Hollis, Eleanor Dixon, and Dennis Newell for their reviews of various reports on which this manuscript is based. This analysis was helped greatly by the input of an external review committee including Steve Webb, Joe Rossabi, Bruce Thomson, Marilyn Gruebel, John Kupar, Malcom Siegel, Mike Smith, and Eliza Frank.

	REFERENCES

TOP ABSTRACT INTRODUCTION SITE DESCRIPTION CONCEPTUAL MODEL NUMERICAL MODEL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Auer, L.A., N.D. Rosenberg, K.H. Birdsell, and E.M. Whitney. 1996. The effects of barometric pumping on contaminant transport. J. Contam. Hydrol. 24:145–166.
• Birdsell, K.H., B.D. Newman, B.A. Robinson, and D.E. Broxton. 2005. Conceptual models of vadose-zone flow and transport beneath the Pajarito Plateau, Los Alamos, New Mexico. Available at www.vadosezonejournal.org. Vadose Zone J. 4:620–636 (this issue).
• Birdsell, K.H., A.V. Wolfsberg, D. Hollis, T.A. Cherry, and K.M. Bower. 2000. Groundwater flow and radionuclide transport calculations for a performance assessment of a low-level waste site. J. Contam. Hydrol. 46:99–129.
• Boas, M.L. 1983. Mathematical methods in the physical sciences. John Wiley & Sons, New York.
• Broxton, D.E., and S.L. Reneau. 1995. Stratigraphic nomenclature of the Bandelier Tuff for the Environmental Restoration Project at Los Alamos National Laboratory. Rep. LA-13010-MS. Los Alamos Natl. Lab., Los Alamos, NM.
• Broxton, D.E., and D.T. Vaniman. 2005. Geologic framework of a groundwater system on the margin of a rift basin, Pajarito Plateau, north-central New Mexico. Available at www.vadosezonejournal.org. Vadose Zone J. 4:522–550 (this issue).
• Conca, J.L., and J. Wright. 1992. Flow and diffusion of unsaturated gravel, soil, and whole rock. Appl. Hydrogeol., Int. Assoc. Hydrol. 1:5–25.
• Fetter, C.W. 1999. Contaminant hydrogeology. Prentice Hall, Englewood Cliffs, NJ.
• Fuentes, H.R., W.L. Polzer, and J.L. Smith. 1991. Laboratory measurements of diffusion coefficients for trichloroethylene and orthoxylene in undisturbed tuff. J. Environ. Qual. 20:215–221.
• George, D. 1997. Unstructured toolbox for modeling and simulation. Rep. LA-UR-97-3052. Los Alamos Natl. Lab., Los Alamos, NM.
• Griggs, R.L. 1964. Geology and groundwater resources of the Los Alamos Area, New Mexico. U.S. Geologic Survey Report to the U.S. Atomic Energy Commission. Water Supply Paper 1753. USGS, Reston, VA.
• Hillel, D. 1982. Introduction to soil physics. Academic Press, Orlando, FL.
• Izett, G.A., J.D. Obradovich. 1994. ⁴⁰Ar/³⁹Ar age constraints for thr Jaramillo Normal Subchron and the Matuyama-Brunhes geomagnetic boundary. J. Geophys. Res. 99(B2):2925–2934. doi:10.1029/93JB03085
• Jury, W.A., W.R. Gardner, and W.A. Gardner. 1991. Soil Physics. Wiley & Sons, NY.
• Jury, W.A., D. Russo, G. Streile, and H. Abd. 1990. Evaluation of volatilization by organic chemicals residing below the soil surface. Water Resour. Res. 26:13–20.
• Kwicklis, E., M. Witkowski, B. Newman, and K. Birdsell. 2005. Development of an infiltration map for the Los Alamos area, New Mexico. Available at www.vadosezonejournal.org. Vadose Zone J. 4:672–693 (this issue).
• LANL. 1994. Ambient monitoring of volatile organic compounds at Los Alamos National Laboratory in TA54, Areas G and L. Rep. LA-SUB-96-99-pt 1. Los Alamos Natl. Lab., Los Alamos, NM.
• LANL. 2002. RFI Report for Material Disposal Area L at Technical Area 54. Rep. LA-UR-02-7803. Los Alamos Natl. Lab., Los Alamos, NM.
• Little, A.D. 1987. The Installation Restoration Program Toxicology Guide. 1:1–16.
• Lyon, B.F., J.A. Holmes, S.T. Kosiewicz, K.A. Wilbert, and C.C. Travis. 1996. Estimation of corrosion on carbon steel TRU waste drums using Poisson distribution. J. Hazard. Mater. 51:165–179.
• McLin, S.G., B.D. Newman, and D.E. Broxton. 2005. Vadose zone characterization and monitoring beneath waste disposal pits using horizontal boreholes, Available at www.vadosezonejournal.org. Vadose Zone J. 4:774–788 (this issue)
• Millington, R.J. 1959. Gas diffusion in porous media. Science (Washington, DC) 130:100–102.
• National Research Council. 2001. Conceptual models of flow and transport in the fractured vadose zone. National Academy Press, Washington, DC.
• Neeper, D.A. 2001. A model of oscillatory transport in granular soils, with application to barometric pumping and earth tides. J. Contam. Hydrol. 48:237–252.
• Neeper, D.A. 2002. Investigation of the vadose zone using barometric pressure cycles. J. Contam. Hydrol. 54:59–80.
• Neeper, D.A., and R.H. Gilkeson. 1996. The influence of topography, stratigraphy, and barometric venting on the hydrology of unsaturated Bandelier Tuff. p. 427–432. In New Mexico Geological Society Guidebook, 47th Field Conference, Jemez Mountains Region. NM Geol. Soc., Las Cruces, NM.
• NIST. 2003. R. Sander, "Henry's Law Constants" in NIST Chemistry WebBook, NIST Standard Reference Database Number 69. In P.J. Linstrom and W.G. Mallard (ed.), March 2003, National Institute of Standards and Technology, Gaithersburg MD, 20899 Available at http://webbook.nist.gov (verified 14 June 2005).
• Ong, S.K., and L.W. Lions. 1991. Effects of soil properties and moisture on the sorption of trichloroehylene vapor. Water Res. 25:29–36.
• Purtymun, W.D., and W.R. Kennedy. 1971. Geology and hydrology of Mesita del Buey: Los Alamos Scientific Laboratory. Rep. LA-4660. Los Alamos Natl. Lab., Los Alamos, NM.
• Reneau, S.L., D.E. Broxton, J.S. Carney, and C. LaDelfe. 1998. Structure of the Tshirege Member of the Bandelier Tuff at Mesita del Buey, Technical Area 54, Los Alamos National Laboratory. Rep. LA-13538-MS. Los Alamos Natl. Lab., Los Alamos, NM.
• Rogers, D.B., and B.M. Gallaher. 1995. The unsaturated hydraulic characteristics of the Bandelier Tuff. Rep. LA-12968-MS. Los Alamos Natl. Lab., Los Alamos, NM.
• Smith, J., B. Hardesty, and N. McCranie. 1999a. Quarterly pore gas sampling at TA-54 MDA L and MDA G First Quarter FY1999, PMC-LA-99-002.
• Smith, J., B. Hardesty, and N. McCranie. 1999b. Quarterly pore gas sampling at TA-54 MDA L and MDA G Second Quarter FY1999, PMC-LA-99-003.
• Smith, J., J. Crocker, and N. McCranie. 1998. Quarterly pore gas sampling at TA-54 MDA L and MDA G Second Quarter FY1998, PMC-LA-98–009.
• Springer, E.P. 2005. Statistical exploration of matrix hydrologic properties for the Bandelier Tuff, Los Alamos, New Mexico. Available at www.vadosezonejournal.org. Vadose Zone J. 4:505–521 (this issue).
• Stauffer, P.H., L.H. Auer, and N.D. Rosenburg. 1997. Compressible gas in porous media: A finite amplitude analysis of natural convection. Int. J. Heat Mass Transfer 40:1585–1589.
• Stauffer, P.H., and N.D. Rosenburg. 1999. Vapor phase transport at a hillside landfill. Environ. Eng. Geosci. 6:61–84.
• Thomson, N.R., J.F. Sykes, and D. Van Vliet. 1997. A numerical investigation into factors affecting gas and aqueous phase plumes in the subsurface, J. Contam. Hydrol. 28:39–70.
• Trease, H., D. George, C.W. Gable, J. Fowler, A. Kuprat, and A. Khamyaseh. 1996. The X3D grid generation system. In B.K. Soni et al. (ed.) Numerical grid generation in computational fluid dynamics and related fields. Engineering Research Center, Mississippi State Univ. Press, Mississippi State, MI.
• Turin, H.J. 1995. Subsurface transport beneath MDA G: A conceptual model. Rep. LA-UR-95-1663. Los Alamos Natl. Lab., Los Alamos, NM.
• van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–898.
• Vaniman, D., G. Cole, J. Gardner, J. Conaway, D. Broxton, S. Reneau, M. Rice, G. WoldeGabriel, J. Blossom, and F. Goff. 1996. Development of a site-wide geologic model for Los Alamos National Laboratory. Rep. LA-UR-00-2059. Los Alamos Natl. Lab., Los Alamos, NM.
• Vogel, T.M., C.S. Criddle, and P.L. McCarty. 1987. Transformation of halogenated aliphatic compounds. Environ. Sci. Technol. 21:722–736.
• Webb, S.W. 1998. Gas-phase diffusion in porous media—Evaluation of an advective-dispersive formulation and the dusty-gas model for binary mixtures. J. Porous Media 1:187–199.
• Weeks, E.P. 2001. Effect of topography on gas flow in unsaturated fractured rock; concepts and observations. p. 53–59. In D.D. Evans et al. (ed.) Flow and transport through unsaturated fractured rock. Geophys. Monogr. 42. AGU, Washington, DC.
• Zyvoloski, G.A., B.A. Robinson, Z.V. Dash, and L.L. Trease. 1997. Summary of the models and methods for the FEHM application—A finite element heat- and mass-transfer code. Rep. LA-13307-MS. Los Alamos Natl. Lab., Los Alamos, NM.

This article has been cited by other articles:

				J. A. Vrugt, P. H. Stauffer, Th. Wohling, B. A. Robinson, and V. V. Vesselinov Inverse Modeling of Subsurface Flow and Transport Properties: A Review with New Developments Vadose Zone J., May 27, 2008; 7(2): 843 - 864. [Abstract] [Full Text] [PDF]

				B. D. Newman and B. A. Robinson The Hydrogeology of Los Alamos National Laboratory: Site History and Overview of Vadose Zone and Groundwater Issues Vadose Zone J., August 16, 2005; 4(3): 614 - 619. [Abstract] [Full Text] [PDF]

				K. H. Birdsell, B. D. Newman, D. E. Broxton, and B. A. Robinson Conceptual Models of Vadose Zone Flow and Transport beneath the Pajarito Plateau, Los Alamos, New Mexico Vadose Zone J., August 16, 2005; 4(3): 620 - 636. [Abstract] [Full Text] [PDF]